Secondary Logo

Journal Logo


Effects of hyperventilation on cerebral oxygen saturation estimated using near-infrared spectroscopy

A randomised comparison between propofol and sevoflurane anaesthesia

Ishiyama, Tadahiko; Kotoda, Masakazu; Asano, Nobumasa; Ikemoto, Kodai; Shintani, Noriyuki; Matsuoka, Toru; Matsukawa, Takashi

Author Information
European Journal of Anaesthesiology: December 2016 - Volume 33 - Issue 12 - p 929-935
doi: 10.1097/EJA.0000000000000507
  • Free



During anaesthesia, cerebral metabolic rate of oxygen (CMRO2) decreases, which in turn reduces cerebral blood flow (CBF). Propofol and sevoflurane are commonly used anaesthetics that act on cerebral vessels. Studies in rats,1 hamsters2 and guinea pigs3 have demonstrated that propofol dilates diaphragmatic, striated muscle and basilar arterioles on exposure to clinically relevant concentrations of propofol. However, in rabbit cerebral pial arterioles, vascular effects were not observed at clinically relevant propofol concentrations.4 Vascular effects of propofol may thus differ among species. Moreover, findings from invasive combined laser Doppler flowmetry and photospectrometry in patients during craniotomies suggest that propofol-induced alterations in CBF and CMRO2 are also regionally specific.5 Nevertheless, accumulating evidence suggests that CBF decreases during propofol anaesthesia in humans.6,7 In contrast, sevoflurane dilates cerebral arterioles in dogs,8 rabbits9 and cats.10 Human studies suggest that global CBF remains unchanged during sevoflurane anaesthesia.11

CBF varies with carbon dioxide partial pressure (PaCO2). Each 1-mmHg increase in PaCO2 increases CBF by 1 to 2 ml 100 g−1 min−1. Changes in CBF associated with alterations of ventilation remain well preserved during both propofol–remifentanil12 and sevoflurane13 anaesthesia, although the magnitude of changes may differ between the anaesthetics.

Near-infrared spectroscopy estimates cerebral regional tissue oxygen saturation (rSO2). Some studies have reported that rSO2 correlates well with jugular bulb oxygen saturation (SjO2)14,15 whereas others found only weak16 or absent correlations.17 Nevertheless, rSO2 is an effective noninvasive tool for monitoring cerebral oxygenation18 and values change as a function of minute ventilation during both propofol and sevoflurane anaesthesia.19

Because the cerebral vascular effects of propofol and sevoflurane anaesthesia may differ, CBF and cerebral oxygenation during hyperventilation may be expected to differ as well. Hyperventilation-induced change in cerebral oxygenation can be evaluated by measurement of rSO2. However, there may be some factors that interfere with rSO2 measurement such as severe diabetes mellitus, renal failure, hepatic failure, etc. Therefore, we decided to include only patients with American Society of Anesthesiologists’ physical status 1 or 2.

We tested two hypotheses: rSO2 changes in response to alterations in ventilation during both propofol and sevoflurane anaesthesia; and the reduction in rSO2 induced by hyperventilation is smaller during sevoflurane anaesthesia than propofol anaesthesia.

Materials and methods

Ethical approval for this study (No. 991) was provided by the Institutional Review Board of University of Yamanashi, Chuo, Yamanashi, Japan (Chairperson Professor Zentaro Yamagata) on 21 November 2012. Written informed consent was obtained from all patients.

We recruited 53 patients (American Society of Anesthesiologists’ physical status 1 or 2) scheduled to undergo elective abdominal surgery. Patients with a known history of cerebral diseases such as cerebral infarction, cerebral haemorrhage, transient ischaemic attack or subarachnoid haemorrhage were excluded. Three patients refused to participate in the study. Therefore, we enrolled 50 patients (Fig. 1). Using a computer-generated list without blocking or stratification, patients were assigned randomly to receive anaesthesia with propofol (propofol; n = 25) or sevoflurane (sevoflurane; n = 25). Sequentially numbered opaque envelopes were opened shortly before induction of anaesthesia; allocation was thus concealed as long as practical.

Fig. 1
Fig. 1:
Study flow diagram.

Two sensors (SAFB-SM, Covidien, Dublin, Ireland) for near-infrared spectroscopy (INVOS 5100C, Covidien, Dublin, Ireland) were pasted on the left and right sides of the forehead with the caudal border approximately 1 cm above the eyebrows to measure rSO2. A bispectral index (BIS) sensor was also pasted on the forehead (Model QUATRO, Covidien, Dublin, Ireland). Propofol was infused by a Diprifusor target-controlled infusion system. (Terufusion TCI Pump TE-371, Terumo, Tokyo, Japan). Propofol plasma target concentration and inhaled sevoflurane concentration were adjusted to maintain BIS between 40 and 60.20 Remifentanil provided analgesia in all patients.

A radial arterial catheter was inserted for continuous monitoring of mean arterial blood pressure (MAP) and blood sampling. Body temperature was monitored by an earphone-type infrared tympanic thermometer (CE Thermo, Nipro, Tokyo, Japan). Our primary measurements (rSO2, SpO2, SaO2, PaO2 and PaCO2) were obtained after induction of anaesthesia, but before surgery started (baseline). Thereafter, hyperventilation to a PaCO2 of approximately 30 mmHg (4 kPa) was induced, and the measurements repeated at 5-min intervals for 15 min. Ventilation was then normalised, and the same set of measurements repeated.

We used Stat Flex version 6.0 (Artec, Osaka, Japan) for statistical analysis. Power analysis indicated that a sample size of 23 patients per group was sufficient to detect a 5% change in rSO2 from the control values with a power of 0.8 and α less than 0.05. Power analysis was based on our pilot study. We applied distribution analysis to assess data distribution. In that analysis, χ2 test was performed, and skewness and kurtosis were calculated. We confirmed that our data were normally distributed. The 95% confidence intervals of rSO2 in the propofol group were 55 to 80% for baseline, 48 to 73% during hyperventilation and 54 to 74% after normalisation of ventilation. The 95% confidence intervals of rSO2 in the sevoflurane group were 56 to 79% for baseline, 48 to 74% during hyperventilation and 53 to 78% after normalisation of ventilation. Within-group comparisons of changes in rSO2, SpO2, SaO2, PaO2 and PaCO2 were examined using analysis of variance and Dunnett post hoc comparisons. Between-group differences in rSO2 were performed using analysis of variance and Newman Keuls post hoc test. Values are represented as mean ± SD. A P value less than 0.05 was considered statistically significant.


Demographic data are shown in Table 1. Age, sex, height and weight were similar in the propofol and sevoflurane groups. MAP increased during hyperventilation (P < 0.01) and then decreased to baseline with normalisation of ventilation in the propofol group (Table 2). MAP remained unchanged in the sevoflurane group. Heart rate and body temperature were similar at all time-points. Propofol plasma target concentration was 2.5 to 2.6 μg ml−1 and end-tidal sevoflurane concentration was 1.11 to 1.25%. Remifentanil doses and BIS values were similar in the propofol and sevoflurane groups (Table 2).

Table 1
Table 1:
Characteristics of patients receiving propofol or sevoflurane
Table 2
Table 2:
Mean arterial blood pressure, heart rate, body temperature, propofol target plasma concentration, end-tidal sevoflurane concentration, dose of remifentanil and bispectral index values during normoventilation and hyperventilation intraoperatively

Baseline PaCO2 was 41 ± 2 mmHg in the propofol group and 42 ± 2 mmHg in the sevoflurane group. During hyperventilation, PaCO2 decreased to 30 ± 2 mmHg in the propofol group and 29 ± 2 mmHg in the sevoflurane group. After normalisation of ventilation, PaCO2 increased to 42 ± 2 mmHg in the propofol group and 42 ± 2 mmHg in the sevoflurane group (Fig. 2).

Fig. 2
Fig. 2:
Changes in arterial PaCO2 tension during HV and NV in the propofol group (a) and the sevoflurane group (b). During hyperventilation, PaCO2 decreased to around 30 mmHg (4 kPa) in both groups. After normalisation of ventilation, PaCO2 was maintained at around 40 mmHg (5.3 kPa). HV, hyperventilation; NV, normoventilation; PaCO2, carbon dioxide partial pressure.

Baseline SpO2 was 99.8 ± 0.5% in the propofol group and 99.4 ± 1.1% in the sevoflurane group. During hyperventilation, SpO2 was 99.9 ± 0.4% in the propofol group and 99.7 ± 0.6% in the sevoflurane group. After normalisation of ventilation, SpO2 was 99.6 ± 0.7% in the propofol group and 99.4 ± 0.8% in the sevoflurane group (Fig. 2). SaO2 was above 98.4% (Fig. 3), and PaO2 was above normal during the experimental period in both groups.

Fig. 3
Fig. 3:
Changes in arterial SaO2 during HV and NV in the propofol group (a) and the sevoflurane group (b). * P < 0.05 compared with 5 min of hyperventilation. HV, hyperventilation; NV, normoventilation; SaO2, oxygen saturation.

The baseline rSO2 values in the propofol group were 67 ± 7% (left) and 67 ± 6% (right), and those in the sevoflurane group were 68 ± 6% (left) and 67 ± 6% (right). There were no differences between the propofol group and the sevoflurane group. During hyperventilation, rSO2 decreased to 60 ± 7% (left, Δ = −10 ± 7%) and 60 ± 6% (right, Δ = −11 ± 8%) in the propofol group (P < 0.01) and to 61 ± 7% (left, Δ = −10 ± 8%) and 60 ± 7% (right, Δ = −9 ± 7%) in the sevoflurane group (P < 0.01). During hyperventilation, seven patients in the propofol group and six patients in the sevoflurane group showed a decrease in rSO2 of 15% or more. Absolute rSO2 values less than 55% were observed in eight patients in the propofol group and nine patients in the sevoflurane group. After normalisation of ventilation, rSO2 rapidly increased to 66 ± 6% (left) and 67 ± 6% (right) in the propofol group and to 68 ± 6% (left) and 67 ± 7% (right) in the sevoflurane group (Fig. 4). Values of rSO2 at each time point did not differ significantly between propofol and sevoflurane.

Fig. 4
Fig. 4:
Changes in cerebral rSO2 during HV and NV in the propofol group (a) and the sevoflurane group (b). rSO2 decreased rapidly during hyperventilation and increased to baseline values along with the normalisation of ventilation. ▪ left side, • right side; * P < 0.05 compared with baseline, ** P < 0.01 compared with baseline. HV, hyperventilation; NV, normoventilation; rSO2, regional oxygen saturation.


The study demonstrated that rSO2 decreased with hyperventilation and remained low during both propofol and sevoflurane anaesthesia. The decrease in rSO2 was comparable in each type of anaesthesia. After the normalisation of PaCO2, rSO2 returned to baseline values. Arterial oxygen saturation remained unchanged throughout the observation period.

Although the adult human brain represents approximately 2% of body weight, the brain consumes about 20% of the entire body's oxygen. The mean decreases in rSO2 were about 10% in both the propofol and sevoflurane groups, and were sustained throughout hyperventilation. In midazolam–fentanyl anaesthesia, rSO2 decreased with hyperventilation (PaCO2 = 30 mmHg) by 7.6%.21 Alexander et al.19 reported that hyperventilation (end-tidal CO2 = 25 mmHg) decreased rSO2 by 9% in propofol–remifentanil anaesthesia and by 10% in sevoflurane anaesthesia. The response to hyperventilation that we observed was thus roughly similar to those which others have reported.

A decrease in rSO2 of 15% or greater during a carotid clamping test increases the risk of severe cerebral ischemia 20-fold.22 Another study indicated that critical threshold values for rSO2 were a 20% or greater decrease from baseline, or an absolute value less than 55% for more than 15 s.23 Previous studies19,21 did not note patients who reached those threshold values. The mean reduction in rSO2 in our patients did not reach these thresholds, although seven patients in the propofol group and six patients in the sevoflurane group had individual reductions in rSO2 of 15% or greater. Furthermore, absolute rSO2 values less than 55% were observed in eight patients in the propofol group and nine patients in the sevoflurane group. Our results thus suggest that even clinically routine levels of hyperventilation may reduce brain oxygenation critically.

Because rSO2 reflects oxygenation of the frontal gyrus, oxygenation in the frontal lobe deteriorated during the hyperventilation period. After traumatic brain injury, the frontal lobes have been shown to be more vulnerable to hypoxia–hypotension than the thalamus in a rat model.24 Although the frontal lobe is critical for personality and higher mental functions, these are not normally assessed in the perioperative context. We must thus consider the possibility that hyperventilation, at least in a subset of patients, may be detrimental in ways that we cannot easily evaluate.

The rSO2 assesses cerebral regional oxygen haemoglobin saturation and reflects the balance between cerebral oxygen supply and oxygen demand. rSO2 in our patients changed rapidly in response to alteration of PaCO2. With hyperventilation, global CBF decreases by 60%25 and CBF in the middle frontal gyrus decreases by 39%.13 However, CMRO2 remains unchanged during hyperventilation.25 With hyperventilation, cerebral oxygen supply thus decreases whereas cerebral oxygen consumption remains unchanged, potentially provoking ischaemia.

Cerebral oxygen saturation measured by near-infrared spectroscopy is reportedly tightly correlated with SjO2 within narrow limits of agreement.15 It has been reported that SjO2 under hyperventilation was lower in propofol anaesthesia than in sevoflurane/N2O anaesthesia in craniotomy patients during intraoperative hypothermia,26 although hypothermia may disturb rSO2 measurement. Others have indicated that the margin of safety against impaired cerebral oxygenation was greater and SjO2 was better preserved with sevoflurane/N2O anaesthesia than with propofol–remifentanil anaesthesia.16 PET studies in healthy humans have shown that propofol reduces regional CBF and regional CMRO2 comparably, whereas sevoflurane reduces regional CBF less than propofol but regional CMRO2 to an extent similar to propofol at a constant hypnotic depth with BIS around 40.27

We thus expected a smaller reduction in rSO2 with sevoflurane than propofol anaesthesia. In fact, the decreases in rSO2 were similar with each anaesthetic. As might be expected, CBF was also influenced by sevoflurane dose. End-tidal sevoflurane was 1.1 to 1.2%, corresponding to a minimum alveolar concentration of 0.6 to 0.7. Previous work indicates that between 0.2 and 1 minimum alveolar concentration of sevoflurane, CBF increases with dose in the insula, but decreases in the frontal gyrus.11 Our results suggest that oxygenation in the frontal gyrus is equally impaired with propofol and sevoflurane anaesthesia during hyperventilation.

MAP increased during hyperventilation in patients given propofol whereas remaining largely unchanged in those assigned to receive sevoflurane. Cerebral autoregulation is preserved during propofol or sevoflurane anaesthesia,28 and MAP values during hyperventilation in the propofol group (80 mmHg) and in the sevoflurane group (68 mmHg) were well within the range of effective autoregulation. Furthermore, cerebral tissue oxygenation has been reported not to be influenced by noradrenaline-induced hypertension during the stable maintenance period of therapeutic hypothermia in comatose cardiac arrest patients when MAP was increased from 70 to 90 mmHg.21 It is therefore unlikely that rSO2 was influenced by MAP.

There are some limitations to this study. First, while we assessed rSO2, we did not measure CBF or SjO2. rSO2 is affected by a variety of factors, including PaO2, temperature and haemoglobin concentration.21 Changes in cerebral perfusion measured by PET are larger than those estimated from rSO2.29 Furthermore, rSO2 can be contaminated by extracerebral blood flow, haemoglobin concentration and the layer of cerebrospinal fluid.30 Although rSO2 is considered a good estimate of frontal lobe oxygenation,31 rSO2 only evaluates a superficial part of the brain. It is not a measure of global CBF. Second, rSO2 measurements were performed at different BIS values (40 to 60). When BIS decreases from 60 to 40, CMRO2 may decrease. Then CBF decreases along with the change of CMRO2. In addition, one study reported that cortical capillary venous oxygen saturation increased with decline in BIS from 40 to 20.5 The rSO2 reflects 25% arterial and 75% venous portion of oxygen.32 Therefore, rSO2 might have been biased by measurements under different BIS values. Third, we only evaluated 15 min of hyperventilation; it is possible that compensatory mechanisms would improve cerebral oxygenation in prolonged hyperventilation.

In summary, in contrast to our expectations, the effects of hyperventilation on rSO2 were similar with propofol and sevoflurane anaesthesia. However, changes in rSO2 correlated with ventilation changes. Our results suggest that even moderate degrees of hyperventilation may critically reduce cerebral oxygenation, especially in patients at high risk of neurocognitive dysfunction or injury from cerebral ischemia and/or hypoxia. However, the risk appears similar with propofol and sevoflurane anaesthesia.

Acknowledgements relating to this article

Assistance with the study: none.

Financial support and sponsorship: support was provided solely from institutional and departmental sources.

Conflicts of interest: none.

Presentation: presented in part at the American Society of Anesthesiologists’ 2014 Annual Meeting, New Orleans, USA, 11 to 15 October 2014.


1. Bazin JE, Dureuil B, Danialou G, et al. Effects of etomidate, propofol and thiopental anaesthesia on arteriolar tone in the rat diaphragm. Br J Anaesth 1998; 81:430–435.
2. Holzmann A, Schmidt H, Gebhardt MM, et al. Propofol-induced alterations in the microcirculation of hamster striated muscle. Br J Anaesth 1995; 75:452–456.
3. Gelb AW, Zhang C, Hamilton JT. Propofol induces dilation and inhibits constriction in guinea pig basilar arteries. Anesth Analg 1996; 83:472–476.
4. Shibuya K, Ishiyama T, Ichikawa M, et al. The direct effects of propofol on pial microvessels in rabbits. J Neurosurg Anesthesiol 2009; 21:40–46.
5. Klein KU, Fukui K, Schramm P, et al. Human cerebral microcirculation and oxygen saturation during propofol-induced reduction of bispectral index. Br J Anaesth 2011; 107:735–741.
6. Song XX, Yu BW. Anesthetic effects of propofol in the healthy human brain: functional imaging evidence. J Anesth 2015; 29:279–288.
7. Vandesteene A, Trempont V, Engelman E, et al. Effect of propofol on cerebral blood flow and metabolism in man. Anaesthesia 1988; 43 (Suppl):42–43.
8. Iida H, Ohata H, Iida M, et al. Isoflurane and sevoflurane induce vasodilation of cerebral vessels via ATP-sensitive K+ channel activation. Anesthesiology 1998; 89:954–960.
9. Ikemoto K, Ishiyama T, Shintani N, et al. The effects of topical and intravenous JM-1232(-) on cerebral pial microvessels of rabbits. BMC Anesthesiol 2015; 15:37.
10. Inoue S, Kawaguchi M, Kurehara K, et al. Mild hypothermia can enhance pial arteriolar vasodilation induced by isoflurane and sevoflurane in cats. Crit Care Med 2002; 30:1863–1869.
11. Schlunzen L, Vafaee MS, Cold GE, et al. Effects of subanaesthetic and anaesthetic doses of sevoflurane on regional cerebral blood flow in healthy volunteers. A positron emission tomographic study. Acta Anaesthesiol Scand 2004; 48:1268–1276.
12. Meng L, Mantulin WW, Alexander BS, et al. Head-up tilt and hyperventilation produce similar changes in cerebral oxygenation and blood volume: an observational comparison study using frequency-domain near-infrared spectroscopy. Can J Anaesth 2012; 59:357–365.
13. Schlünzen L, Vafaee MS, Juul N, et al. Regional cerebral blood flow responses to hyperventilation during sevoflurane anaesthesia studied with PET. Acta Anaesthesiol Scand 2010; 54:610–615.
14. Nagdyman N, Ewert P, Peters B, et al. Comparison of different near-infrared spectroscopic cerebral oxygenation indices with central venous and jugular venous oxygenation saturation in children. Paediatr Anaesth 2008; 18:160–166.
15. Yoshitani K, Kuwajima K, Irie T, et al. Clinical validity of cerebral oxygen saturation measured by time-resolved spectroscopy during carotid endarterectomy. J Neurosurg Anesthesiol 2013; 25:248–253.
16. Jeong H, Jeong S, Lim HJ, et al. Cerebral oxygen saturation measured by near-infrared spectroscopy and jugular venous bulb oxygen saturation during arthroscopic shoulder surgery in beach chair position under sevoflurane-nitrous oxide or propofol-remifentanil anesthesia. Anesthesiology 2012; 116:1047–1056.
17. Ali MS, Harmer M, Vaughan RS, et al. Spatially resolved spectroscopy (NIRO-300) does not agree with jugular bulb oxygen saturation in patients undergoing warm bypass surgery. Can J Anaesth 2001; 48:497–501.
18. de Vries JW, Visser GH, Bakker PF. Neuromonitoring in defibrillation threshold testing. A comparison between EEG, near-infrared spectroscopy and jugular bulb oximetry. J Clin Monit 1997; 13:303–307.
19. Alexander BS, Gelb AW, Mantulin WW, et al. Impact of stepwise hyperventilation on cerebral tissue oxygen saturation in anesthetized patients: a mechanistic study. Acta Anaesthesiol Scand 2013; 57:604–612.
20. Escallier KE, Nadelson MR, Zhou D, et al. Monitoring the brain: processed electroencephalogram and peri-operative outcomes. Anaesthesia 2015; 69:899–910.
21. Bouzat P, Suys T, Sala N, et al. Effect of moderate hyperventilation and induced hypertension on cerebral tissue oxygenation after cardiac arrest and therapeutic hypothermia. Resuscitation 2013; 84:1540–1545.
22. Rigamonti A, Scandroglio M, Minicucci F, et al. A clinical evaluation of near-infrared cerebral oximetry in the awake patient to monitor cerebral perfusion during carotid endarterectomy. J Clin Anesth 2005; 17:426–430.
23. Murphy GS, Szokol JW, Marymont JH, et al. Cerebral oxygen desaturation events assessed by near-infrared spectroscopy during shoulder arthroscopy in the beach chair and lateral decubitus positions. Anesth Analg 2010; 111:496–505.
24. Blanie A, Vigue B, Benhamou D, et al. The frontal lobe and thalamus have different sensitivities to hypoxia-hypotension after traumatic brain injury: a microdialysis study in rats. J Neurotrauma 2012; 29:2782–2790.
25. Grüne F, Kazmaier S, Sonntag H, et al. Moderate hyperventilation during intravenous anesthesia increases net cerebral lactate efflux. Anesthesiology 2014; 120:335–342.
26. Kawano Y, Kawaguchi M, Inoue S, et al. Jugular bulb oxygen saturation under propofol or sevoflurane/nitrous oxide anesthesia during deliberate mild hypothermia in neurosurgical patients. J Neurosurg Anesthesiol 2004; 16:6–10.
27. Kaisti KK, Långsjö JW, Aalto S, et al. Effects of sevoflurane, propofol, and adjunct nitrous oxide on regional cerebral blood flow, oxygen consumption, and blood volume in humans. Anesthesiology 2003; 99:603–613.
28. Dagal A, Lam AM. Cerebral autoregulation and anesthesia. Curr Opin Anaesthesiol 2009; 22:547–552.
29. Rostrup E, Law I, Pott F, et al. Cerebral hemodynamics measured with simultaneous PET and near-infrared spectroscopy in humans. Brain Res 2002; 954:183–193.
30. Yoshitani K, Kawaguchi M, Miura N, et al. Effects of hemoglobin concentration, skull thickness, and the area of the cerebrospinal fluid layer on near-infrared spectroscopy measurements. Anesthesiology 2007; 106:458–462.
31. Rasmussen P, Dawson EA, Nybo L, et al. Capillary-oxygenation-level-dependent near-infrared spectrometry in frontal lobe of humans. J Cereb Blood Flow Metab 2007; 27:1082–1093.
32. Pollard V, Prough DS, DeMelo AE, et al. Validation in volunteers of a near-infrared spectroscope for monitoring brain oxygenation in vivo. Anesth Analg 1996; 82:269–277.
© 2016 European Society of Anaesthesiology